AI + DePin Project Review

In this section, we will explore several AI-related DePin projects, focusing on the decentralized file storage and access platform Filecoin, the decentralized GPU computing power rental platform Io.net, and the decentralized AI model deployment and access platform Bittensor. These three play an important role in data storage access, computing power support training, and model deployment in the field of AI.

Filecoin

Filecoin is a decentralized storage network that enables distributed data storage worldwide through blockchain technology and cryptocurrency economic models. Developed by Protocol Labs, Filecoin aims to create an open and public storage market where users can purchase storage space in the network by paying Filecoin tokens (FIL) or earn FIL by providing storage services.

Function

• Decentralized storage: Filecoin stores data in a decentralized manner, avoiding the centralized drawbacks of traditional cloud storage, such as single point failure and data censorship risks.

• Market-driven: Filecoins storage market is determined by supply and demand. Storage prices and service quality are dynamically adjusted through free market mechanisms. Users can choose the best storage solution based on their needs.

• Verifiable storage: Filecoin ensures that data is effectively stored and backed up at the storage provider through mechanisms such as Proof-of-Spacetime (PoSt) and Proof-of-Replication (PoRep).

• Incentive mechanism: Through mining and transaction reward mechanisms, Filecoin encourages network participants to provide storage and retrieval services, thereby increasing the storage capacity and availability of the network.

• Scalability: The Filecoin network supports large-scale data storage and fast access by introducing technical means such as sharding to meet the needs of massive data growth in the future.

Pain points solved

• High data storage costs: Through Filecoins decentralized storage market, users can choose storage providers more flexibly and reduce data storage costs.

• Data security and privacy issues: Decentralized storage and encryption technology ensure the privacy and security of data, reducing the risk of data leakage caused by centralized storage.

• Reliability of data storage: The space-time proof and replication proof mechanisms provided by Filecoin ensure the integrity and verifiability of data during the storage process, improving the reliability of data storage.

• Trust issues in traditional storage platforms: Filecoin achieves storage transparency through blockchain technology, eliminates the monopoly and manipulation of data by third-party institutions, and enhances users trust in storage services.

Target Users

• Storage providers: respond to users storage requests and earn tokens by providing idle disk space access to the platform. Storage providers need to stake tokens. If they fail to provide valid storage proofs, they will be punished and lose some of their staked tokens.

• File Retriever: When a user needs to access a file, they can earn tokens by retrieving the file location. File Retrievers do not need to stake tokens.

• Data storer: Through the market mechanism, submit the price they are willing to pay, and after matching with the storer, send the data to the storer. Both parties sign the transaction order and submit it to the blockchain.

• Data user: The user submits a unique file identifier and pays a price, and the file retriever will find the storage location of the file, respond to the storage request and provide the data.

Token Economic System

• Circulation of FIL tokens: FIL is the native cryptocurrency in the Filecoin network, used to pay for storage, reward miners, and conduct transactions in the network. The circulation of FIL tokens maintains the normal operation of the Filecoin network.

• Rewards for storage miners and retrieval miners: Storage providers earn FIL tokens by providing storage space and data retrieval services. Miners’ rewards are related to the storage space they provide, the frequency of data access, and their contribution to network consensus.

• Network fees: Users need to pay FIL tokens to purchase storage and retrieval services. The fees are determined by the supply and demand of the storage market. Users can freely choose suitable service providers in the market.

• Token issuance and inflation: The total supply of Filecoin is 2 billion, and new FIL tokens are gradually issued through mining rewards. As the number of miners increases, the inflation rate of the network will gradually decrease.

Io.net

Io.net is a distributed GPU computing platform that collects and clusters idle computing power to provide computing power scheduling and temporary supplementation to the market, rather than replacing existing cloud computing resources. The platform allows suppliers to deploy supported hardware for users to rent through simple Docker instructions to meet the needs of task distribution and processing. Through the model of distributed computing power sharing, Io.net hopes to provide effects close to those of cloud computing platforms while significantly reducing service costs.

Function

• Easy deployment: Suppliers can easily deploy hardware through Docker instructions, and users can conveniently rent hardware clusters through the platform to obtain the required computing power.

• Clustered computing power: By clustering idle computing power, the platform acts as a dispatcher and temporary supplement of market computing power, thereby improving the overall utilization of computing resources.

• Secure transmission and on-chain storage: The platform uses end-to-end encryption technology to ensure the security of user data. At the same time, task execution information will be stored on the chain to achieve transparent and permanent storage of logs.

• Node health monitoring: The platform records and discloses the health status of each node, including offline time, network speed, and task execution status, to ensure the stability and reliability of the system.

Pain points solved

• Insufficient computing power: Due to the rise of large models, the market demand for GPU computing power required for training has increased dramatically. Io.net fills this computing power gap by integrating idle GPU resources from the public.

• Privacy and Compliance: Large cloud platform service providers such as AWS and Google Cloud have strict KYC requirements for users, while Io.net avoids compliance issues through decentralization, allowing users to choose to use resources more flexibly.

• High cost: The service prices of cloud computing platforms are relatively high, but Io.net significantly reduces costs through distributed computing power sharing, while achieving service quality close to that of cloud platforms through clustering technology.

Target Users

• Computing power providers: connect idle GPUs to the platform for others to use. Based on the performance and stability of the equipment provided, you can get token rewards.

• Computing power users: rent GPUs or GPU clusters by consuming tokens for task submission or large model training.

• Pledgers: Pledgers pledge platform tokens to support the long-term stable operation of the platform and obtain pledge income from equipment leasing, which helps to improve the ranking of excellent equipment.

Token Economic System

• Token usage: All transactions within the platform use the native token $IO to reduce transaction friction in smart contracts. Users and suppliers can pay with USDC or $IO, but using USDC requires a 2% service fee.

• Total token supply: $IO has a maximum supply of 800 million, 500 million will be issued at launch, and the remaining 300 million will be used to reward suppliers and stakers. The tokens will be gradually released over 20 years, starting with 8% of the total in the first year and decreasing by 1.02% per month.

• Token Burning: A portion of the platform revenue will be used to buy back and burn $IO, with the cost coming from a 0.25% bilateral reservation fee and a 2% service fee for payments made using USDC.

• Token distribution: Tokens will be distributed to seed investors, Series A investors, team, ecosystem and community, and supplier rewards.

Bittensor (TAO)

Bittensor is a decentralized peer-to-peer AI model market that aims to promote the production and circulation of AI models by allowing different intelligent systems to evaluate and reward each other. Through a distributed architecture, Bittensor has created a market that can continuously produce new models and reward contributors for the value of their information. The platform provides researchers and developers with a platform to deploy AI models to earn revenue, while users can use various AI models and functions through the platform.

Function

• Distributed Marketplace: Bittensor has built a decentralized AI model marketplace, allowing engineers and small AI systems to monetize their work directly, breaking the monopoly of large companies on AI.

• Standardization and modularity: The network supports multiple modes (such as text, images, and voice), allowing different AI models to interact and share knowledge, and can be expanded to more complex multimodal systems.

• System ranking: Each node is ranked according to its contribution to the network. The contribution measurement criteria include the performance of the node on the task, the evaluation of its output by other nodes, and the trust it has gained in the network. Nodes with higher rankings will receive more network weight and rewards, which motivates nodes to continue to provide high-quality services in the decentralized market. This ranking mechanism not only ensures the fairness of the system, but also improves the overall computing efficiency and model quality of the network.

Pain points solved

• Centralization of intelligent production: The current AI ecosystem is concentrated in a few large companies, making it difficult for independent developers to monetize. Bittensor provides independent developers and small AI systems with direct profit opportunities through a peer-to-peer decentralized market.

• Low utilization of computing resources: Traditional AI model training relies on a single task and cannot fully utilize diverse intelligent systems. Bittensor allows different types of intelligent systems to collaborate with each other and improve the utilization efficiency of computing resources.

Target Users

• Node operators: connect computing power and models to the Bittensor network, and receive token rewards by participating in task processing and model training. Node operators can be independent developers, small AI companies, or even individual researchers, who can improve their ranking and income in the network by providing high-quality computing resources and models.

• AI model users: Users who need AI computing resources and model services rent computing power and intelligent models in the Bittensor network by paying tokens. Users can be enterprises, scientific research institutions or individual developers who use high-quality models in the network to complete specific tasks, such as data analysis, model reasoning, etc.

• Stakers: Users holding Bittensor tokens support the long-term stable operation of the network through staking and receive staking rewards. Stakers can not only benefit from the inflation of the network, but also improve the ranking of the nodes they support through staking, thereby indirectly affecting the overall computing efficiency and income distribution of the network.

Token Economic System

• Token usage: All transactions and incentives within the Bittensor network are conducted through native tokens, reducing friction in the transaction process. Users can use tokens to pay for computing resources and model services, and node operators earn tokens by providing services.

• Token generation: Every 12 seconds, a block is generated, generating 1 TAO token, which is distributed based on the performance of the subnet and the performance of the nodes in it. The distribution ratio of tokens is: 18% is allocated to the subnet owner, and the subnet miners and validators each receive 41%. The maximum supply of tokens is 21 million.

Challenges and conclusions of DePin

As an emerging network architecture, DePIN realizes the decentralized management of physical infrastructure by combining blockchain technology. This innovation not only solves the problems of data privacy, service interruption and high expansion cost faced by traditional infrastructure, but also gives network participants more control and participation through token incentive mechanism and self-organization model. Although DePIN has shown great potential, it still faces some challenges.

• Scalability: DePIN’s scalability problem stems from its reliance on the decentralized nature of blockchain technology. As the number of users and network size increase, the transaction volume on the blockchain network will also increase. In particular, the connection between DePIN applications and the physical world requires higher information transmission requirements. This will lead to longer transaction confirmation times and increased transaction fees, which in turn affects the overall network efficiency and user experience.

• Interoperability: The DePIN ecosystem is built on multiple blockchains, which requires DePIN applications to support homogeneous or heterogeneous state transitions and achieve seamless interoperability with other blockchain networks. However, current interoperability solutions are usually limited to specific blockchain ecosystems or accompanied by high cross-chain costs, making it difficult to fully meet the needs of DePIN.

• Regulatory compliance: As part of the Web 3.0 ecosystem, DePIN faces multiple regulatory challenges. Its decentralized and anonymous nature makes it difficult for regulators to monitor the flow of funds, which may lead to an increase in illegal fundraising, pyramid schemes, and money laundering. In addition, in terms of tax supervision, due to the anonymity of accounts, it is difficult for the government to collect evidence required for taxation, which poses a challenge to the existing tax system.

In the future, the development of DePIN will depend on the solution of these key issues, and is expected to play an important role in a wide range of application scenarios and reshape the operation mode of physical infrastructure.